| name | sessions |
| description | Evidence-first analysis of local AI-agent session history with Sessions. Use when past sessions are needed for context recovery or transfer, retrospectives, preference discovery, skill or workflow audits, verification audits, handoff continuity, or reusable-capability discovery. |
Sessions
Turn retained local agent sessions into a traceable answer. Preserve what the
evidence proves, what it does not prove, and how another person can reproduce the
result.
Analyze
- Route. Choose the one route that best matches the user's primary decision.
| User goal | Playbook |
|---|
| Recover a decision, prior research, or transferable context | Search and context |
| Explain failure, drift, reversals, recovery, or unresolved work | Retrospective |
| Find repeated working or communication preferences | Preferences |
| Audit whether a skill or workflow was eligible, used, and followed | Workflow audit |
| Compare completion claims with recorded checks and later corrections | Verification audit |
| Check what a parent, child, fork, or continuation transferred | Handoff continuity |
| Find recurring work or friction worth a reusable capability | Capability discovery |
Done when one primary outcome and playbook are selected. Add a second route
only when the first result leaves a required question unresolved.
-
Load. Read the evidence protocol, then
only the selected playbook.
Done when both completion bars are available.
-
Investigate. Use the smallest useful sessions queries, then report facts,
interpretation, limits, and the next action.
Done when the question is answered with reproducible canonical evidence, or
the exact missing evidence or authorization is named.
